1448
Langmuir 2005, 21, 1448-1456
Reduction of NO Adlayers on Pt(110) and Pt(111) in Acidic Media: Evidence for Adsorption Site-Specific Reduction Victor Rosca,* Guillermo L. Beltramo, and Marc T. M. Koper Schuit Institute of Catalysis, Laboratory of Inorganic Chemistry and Catalysis, Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands Received September 30, 2004 We present a combined in situ Fourier transform infrared reflection-absorption spectroscopy and voltammetric study of the reduction of saturated and subsaturated NO adlayers on Pt(111) and Pt(110) single-crystal surfaces in acidic media. The stripping voltammetry experiments and the associated evolution of infrared spectra indicate that different features (peaks) observed in the voltammetric profile for the electrochemical reduction of NO adlayers on the surfaces considered are related to the reduction of NOads at different adsorption sites and not to different (consecutive) processes. More specifically, reduction of high- and intermediate-coverage (ca. 0.5-1 monolayers (ML)) NO adlayers on Pt(110) is accompanied by site switching from atop to bridge position, in agreement with the ultra-high-vacuum data. On Pt(111) linearly bonded (atop) NO and face-centered cubic 3-fold-hollow NO species coexist at high coverages (0.25-0.5 ML) and can be reduced consecutively and independently. On Pt(111) and Pt(110) electrodes, linearly bonded NO species are more reactive than multifold-bonded NO species. Both spectroscopic and voltammetric data indicate that ammonia is the main product of NOads reduction on the two surfaces examined.
1. Introduction The adsorption and reaction of nitric oxide (NO) on transition metal surfaces in an electrochemical environment are of considerable technological and scientific interest. NO is a key intermediate or reagent in a series of environmentally and industrially important processes, such as nitrate (nitrite) reduction and ammonia oxidation1 or the liquid-phase catalytic hydrogenation of nitric oxide (NO) or nitrate to hydroxylamine.2,3 Furthermore, electrochemical sensing and removal of nitric oxide4 requires a thorough understanding of electrochemistry of NO. From a fundamental point of view, NO reduction is a relatively simple electrocatalytic process and could serve as a model system for studying activity and selectivity issues in electrocatalysis. Finally, similar to the extensively studied oxidation of carbon monoxide, NO is in many respects a suitable molecule for comparing reactions at the solidgas and solid-liquid interface, especially when combined with work at single-crystal surfaces. Recent studies of the electrochemical transformation of NO adlayers (NOads) on low-index single-crystal platinum electrodes5-13 have significantly improved our un(1) Horold, S.; Vorlop, K.-D.; Tacke, T.; Sell, M. Catal. Today 1993, 17, 21. (2) Ritz, J.; Fuchs, H.; Perryman, H. G. Hydroxylamine. In Ulmann’s Encyclopedia of Industrial Chemistry, 6th ed.; Wiley: Chichester, 2000. (3) van de Moesdijk, C. G. M. The catalytic reduction of nitrate and nitric oxide to hydroxylamine: kinetics and mechanism. Ph.D. Thesis, Eindhoven University of Technology, 1979. (4) Allen, B. W.; Piantadosi, C. A.; Coury, L. A., Jr. Nitric Oxide 2000, 4, 75. (5) Ye, S.; Hattori, H.; Kita, H. Ber. Bunsen-Ges. Phys. Chem. 1992, 73, 1884. (6) Ye, S.; Kita, H. J. Electroanal. Chem. 1993, 346, 489. (7) Rodes, A.; Gomez, R.; Orts, J. M.; Feliu, J. M.; Perez, A.; Aldaz, A. J. Electroanal. Chem. 1993, 359, 315. (8) Rodes, A.; Gomez, R.; Orts, J. M.; Feliu, J. M.; Perez, A.; Aldaz, A. Langmuir 1995, 11, 3549. (9) Rodes, A.; Gomez, R.; Perez, A.; Feliu, J. M.; Aldaz, A. Electrochim. Acta 1996, 41, 729. (10) Momoi, K.; Song, M. B.; Ito, M. J. Electroanal. Chem. 1999, 473, 43. (11) Zang, Z.-H.; Wu, Z.-L.; Yau, S.-L. J. Phys. Chem. B 1999, 103, 9624.
derstanding of the reactivity of nitric oxide, as compared to the knowledge gained from studies on polycrystalline platinum.14 Recently, we have performed a systematic voltammetric analysis of reduction and oxidation of NO adlayers on Pt(111), Pt(110), and a series of stepped Pt[n(111) × (111)] electrodes,15 aimed at establishing both the mechanism and structure sensitivity of reactions of adsorbed NO. The following tentative mechanism was proposed for the reduction of adsorbed NO:15
NOads + H+ + e- / HNOads
equilibrium (1)
HNOads + H+ + e- f H2NOads H2NOads + 4H+ + 3e- f NH4+ + H2O
rds fast
(2) (3)
Most importantly, the first two reaction steps are assumed to be two concerted proton/electron transfers. It is difficult to elaborate on the nature and the elementary steps of the process (3), since they take place after the rate-determining step. Reduction of NOads appears to be structure insensitive with regard to the kinetics of the process: the NO adlayer may be stripped off in two to three voltammetric features in the potential region between ca. 0.15 and 0.4 V vs RHE (0.5 M sulfuric or perchloric acid) at all surfaces examined. On the other hand, the voltammetric profile was observed to be clearly structure sensitive. This structure sensitivity was ascribed to different NO surface configurations and coverages, combined with structure sensitive adsorption of hydrogen and anions, rather than with a structure sensitivity of the NO reduction per se.15 (12) Casero, E.; Alonso, C.; Martin-Gago, J. A.; Bogatti, F.; Felici, R.; Renner, R.; Lee, T.; Zegenhagen, J. Surf. Sci. 2002, 507-510. (13) Gomez, R.; Rodes, A.; Orts, J. M.; Feliu, J. M.; Perez, J. M. Surf. Sci. 1995, 342, L1104. (14) de Vooys, A. C. A.; Beltramo, G. L.; van Riet, B.; van Veen, J. A. R.; Koper, M. T. M. Electrochim. Acta 2004, 49, 1307. (15) Beltramo, G. L.; Koper, M. T. M. Langmuir 2003, 19, 8907.
10.1021/la0475831 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/20/2005
Reduction of NO Adlayers on Pt
Since we aim at a fundamental (molecular level) understanding of NO electrochemical reduction, we combine electrochemical techniques, which give a macroscopic view of the system, with nonelectrochemical molecular level probes, such as infrared spectroscopy. Due to its ability to supply vibrational information and its in situ applicability to electrochemical interfaces, Fourier transform infrared reflection-absorption spectroscopy (FTIRRAS) has become an important tool for investigating electrode processes at a molecular level.16 An important, if not critical, factor enabling FTIRRAS application to in situ electrochemical studies was the development of links between related metal-vacuum and metal-solution interfaces. In this sense, in addition to extensive studies on carbon monoxide (CO), adsorption of NO on platinumgroup metals has been receiving increasing attention in recent years.13,17-19 Described herein is a combined in situ FTIRRAS and voltammetric study of the reduction of saturated and subsaturated NO adlayers on Pt(111) and Pt(110) single crystal surfaces in acidic media. We aim particularly at the structure sensitivity of the voltammetric profile in relation to the coverage and the associated evolution of FTIRRAS spectra. Specifically, we shall demonstrate a relationship between the NO binding site(s) and the voltammetric peak(s) on the surfaces studied. 2. Experimental Section The H2SO4 and HClO4 working solutions were prepared from their respective concentrated acids (“Suprapur”, Merck) and ultrapure water (Millipore MilliQ system, 18.2 MΩ cm, less than 3 ppb total organic carbon) or triply distilled deuterium oxide (Merck, 99.8%). Two types of single-crystal platinum electrodes were used. Bead-type single-crystal platinum electrodes, prepared by Clavilier’s method,20 were used for most of the electrochemical measurements. For in situ FTIR measurements, commercial platinum disk electrodes of 10 mm diameter were used (oriented within 1°, Surface Preparation Laboratory, Zaandam, The Netherlands). Prior to each experiment, the working electrode was flame-annealed, cooled to room temperature in an Ar:H2 (3:1) atmosphere, and transferred to the electrochemical cell under the protection of a droplet of deoxygenated ultrapure water. To avoid the so-called missing row reconstruction, the Pt(110) electrode was cooled in an intensive stream of Ar, following the approach described by Markovic et al.21 Electrochemical measurements were performed in a singlecompartment three-electrode glass cell, using a computercontrolled potentiostat (AutoLab-PGSTAT20, Eco Chemie, Utrecht, The Netherlands). The cell contained a small movable glass spoon22 that allowed dosing of NO at open circuit potential from a diluted NaNO2 solution under the argon atmosphere, thus allowing deposition of saturated and subsaturated NO adlayers without dissolving NO in the working solution. In the case of in situ FTIR experiments, NO adlayers were generated in a similar way: immersion of the working electrode in a separate solution of NaNO2 was followed by a thorough washing of the electrode with pure deuterated water and subsequent transfer of the electrode to the cell. The cell and the other glassware were cleaned by boiling in a 1:1 mixture of concentrated nitric and sulfuric acid, followed by repeated boiling with ultrapure water. A coiled (16) Iwasita, T.; Nart, F. C. Prog. Surf. Sci. 1997, 55, 271. (17) Villegas, I.; Gomez, R.; Weaver, M. J. J. Phys. Chem. 1995, 99, 14832. (18) Weaver, M. J.; Zou, S.; Tang, C. J. Chem. Phys. 1999, 111, 368. (19) Zou, S.; Gomez, R.; Weaver, M. J. J. Electroanal. Chem. 1999, 474, 155. (20) Clavilier, J.; Armand, D.; Sun, S. G.; Petit, M. J. Electroanal. Chem. 1986, 205, 267. (21) Markovic, N. M.; Grgur, B. N.; Lucas, C. A.; Ross, P. N. Surf. Sci. 1997, 384, L805. (22) Feliu, J. M.; Orts, J. M.; Fernandez-Vega, A.; Aldaz, A.; Clavilier, J. J. Electroanal. Chem. 1990, 296, 191.
Langmuir, Vol. 21, No. 4, 2005 1449
Figure 1. Cyclic voltammograms for the reduction of NO adlayers at different initial coverage (as indicated by the numbers in the legend) on Pt(110) in 0.5 M H2SO4 at 2 mV s-1. The solid thick line is the voltammogram for the reduction of a saturated NO adlayer, while the solid thin line is the Pt(110) blank voltammetry. The inset shows the voltammetry for the reduction of a saturated NO adlayer (the solid line) on Pt(110) and the so-called blank voltammetric profile (the dotted line) in 0.5 M HClO4 at 2 mV s-1. platinum wire served as counter electrode. In HClO4 solutions, the reference electrode was an internal reversible hydrogen electrode (RHE). In H2SO4 solutions a saturated mercurymercury sulfate (Hg/Hg2SO4/K2SO4) electrode, connected via a Luggin capillary, was used as a reference. However, all potentials are quoted versus the RHE. Prior to each experiment, all solutions were deoxygenated by purging with pure (N50) argon. The in situ FTIRRAS measurements were performed under external reflection conditions. All the spectra reported were recorded under potential control. The Fourier transform infrared spectrometer was a Bru¨ker IFS113V, equipped with a narrowband MCT detector. The design of the spectroelectrochemical cell closely resembles that described elsewhere.23 The cell featured a prismatic CaF2 transmission window beveled at 60°. Five hundred interferograms were collected at each potential. The spectral resolution was 8 cm-1. The reflectance spectra were calculated as (R - R0)/R0, where R and R0 are the reflectance at the sample and the reference potential, respectively. Consequently, the ∆R/R ratio gives negative bands for species that are formed and positive bands for species that are consumed at the sample potential, as compared to the reference potential. The surface order and cleanliness of the electrodes were checked before each experiment by recording the so-called blank cyclic voltammograms and their comparison to standard voltammetric profiles.24-26 Specifically, the voltammetric profile for the Pt(110) electrode (Figure 1) agrees well with those reported by other groups,21,27,28 allowing us to conclude that our electrode preparation procedure results in a well-ordered Pt(110)-(1 × 1) surface (for a detailed discussion see ref 26).
3. Results and Data Analysis 3.1. Pt(110). We shall begin with presenting the voltammetric data for the reductive stripping of NO adlayers on Pt(110). Figure 1 shows the stripping voltammetry of saturated and subsaturated NO adlayers on (23) Iwasita, T.; Nart, F. C.; Vielstich, W. Ber. Bunsen-Ges. Phys. Chem. 1990, 94, 1030. (24) Lebedeva, N. P.; Koper, M. T. M.; Herrero, E.; Feliu, J. M.; van Santen, R. A. J. Electroanal. Chem. 2000, 487, 37. (25) Lebedeva, N. P.; Rodes, A.; Feliu, J. M.; Koper, M. T. M.; van Santen, R. A. J. Phys. Chem. B 2002, 106, 9863. (26) Lebedeva, N. P.; Koper, M. T. M.; Feliu, J. M.; van Santen, R. A. J. Phys. Chem. B 2002, 106, 12938. (27) Gomez, R.; Clavilier, J. J. Electroanal. Chem. 1993, 354, 189. (28) Kibler, L. A.; Cuesta, A.; Kleinert, M.; Kolb, D. M. J. Electroanal. Chem. 2000, 484, 73.
1450
Langmuir, Vol. 21, No. 4, 2005
Pt(110) at low scan rate. Note that the subsaturated NO adlayers were deposited by immersion in NaNO2 solution for different periods of time (see the Experimental Section for details). The voltammetric profile for the reduction of a saturated NO adlayer in sulfuric acid exhibits three characteristic peaks: E1110 at ca. 0.33 V, E2110 at ca. 0.23 V, whereas the third reduction peak (E3110) is assumed to correspond mainly to the hydrogen underpotential deposition (Hupd), since (i) the charge associated with this peak is similar to the charge for the Hupd/(bi)sulfate adsorption/ desorption on clean Pt(110) and (ii) the position and charge of this peak are only slightly affected by NO initial coverage. The shift of the position of the peak E3110, as compared to a clean surface, and the apparent irreversibility (Tafel slope of 32 mV decade-1) is most probably related to some residual NO left on the surface. The inset of Figure 1 shows the stripping voltammetry for the reduction of a saturated NO adlayers in perchloric acid (continuous line). In perchloric acid the main reduction peak involves more charge than in sulfuric acid, while the peaks E1110 and E2110 significantly overlap. Note that in perchloric acid both reduction peaks are positioned slightly more positively than those in sulfuric acid, thus indicating an effect of the (bi)sulfate coadsorption for experiments in sulfuric acid. As argued elsewhere,15 it would be reasonable to assume that in perchloric acid practically all adsorbed NO is stripped off at ca. 0.2 V, whereas in sulfuric acid a small amount of NO may be still present at this potential. Figure 1 also shows that the peak E3110 does not exhibit a significant change in charge with the initial NO coverage, which is in agreement with assigning this feature to the Hupd/(bi)sulfate adsorption/desorption. Nevertheless, the peak position shifts slightly negatively with increasing NO coverage. The charge corresponding to the first two reduction peaks increases with increasing NO coverage, while their positions remain the same. Importantly, the reduction peak E2110 is “saturated” at intermediate NO coverage, while the charge under the main peak (E1110) increases until the maximum NO coverage (ca. 0.9-1)15 is reached. This observation suggests that the two reduction features are rather due to NO reduction at different adsorption sites than due to different (consecutive) processes. If the voltammetric peaks E1110 and E2110 are due to reduction of NO molecules at different adsorption sites, it would be interesting to see if they may be reduced independently. The experimental results shown in Figure 2 indicate that this is indeed the case. Three voltammetric runs were performed, starting with 0.5 V and recorded one after another, with decreasing the lower potential limit (as indicated by the numbers in the figure). Note that the voltammetric features cannot be reduced completely independently, since they partially overlap. Significantly, the position and charge of voltammetric peak E3110 is not affected by this experiment as compared to the profile for an uninterrupted stripping of a saturated NO adlayer (Figure 2, the thick continuous line). This result is in good agreement with our interpretation of this feature given above. Finally, the same experiment in perchloric acid gives the same result: the voltammetric peaks E1110 and E2110, attributed to reduction of adsorbed NO at different adsorption sites, can be reduced consecutively and independently. Turning now to the spectroscopic data for NO reduction on Pt(110), Figure 3 shows a series of spectra for a (initially) saturated NO adlayer on Pt(110) electrode. To obtain unipolar (better-defined) bands, the reference potential was chosen at 0.1 V, i.e., the potential where NO is stripped
Rosca et al.
Figure 2. Cyclic voltammograms showing the independent stripping of the voltammetric features characteristic to NO reduction on Pt(110) (see text for details). Experimental conditions: 0.5 M H2SO4; 2 mV s-1. Numbers in the legend indicate the lower potential limit of the corresponding CVs. The solid thick line is voltammogram for reduction of a saturated NO adlayer, while the solid thin line is the Pt(110) blank voltammetry.
Figure 3. The potential difference infrared spectra acquired at a Pt(110) electrode covered with a preadsorbed NO adlayer. Numbers indicate the applied electrode potential on the RHE scale. Experimental conditions: 0.1 M HClO4 (in D2O), reference potential 0.1 V, p-polarized light.
off completely, as deduced from the CV data (see above). In the potential region in which chemisorbed NO is stable (between ca. 0.7 and 0.45 V), the spectra show two characteristic absorption bands: a broad band between ca. 1800 and 1680 cm-1 with the maximum contribution around 1770 cm-1 and a weaker band at around 1600 cm-1. The two bands are not present in spectra acquired by using s-polarized light under the same experimental conditions, thus indicating that the two bands originate
Reduction of NO Adlayers on Pt
from adsorbed species, namely, adsorbed NO. At the same time, the two spectroscopic features mentioned show a weak, though significant, dependence of the applied potential. Accordingly, in the potential region of NO stability the maximum of both bands slightly shifts toward lower wavenumbers with decreasing applied potential (see Figure 3). This trend is expected for N-O stretching (νNO) in adsorbed NO species.18 Before discussing the interpretation of the spectra in Figure 3, let us note that the resulting spectroscopic picture is strikingly similar to that for NO adsorption at the Pt(110)/ultrahigh vacuum (UHV) interface. Here we shall refer to a recent study by Brown and co-workers.29 These authors analyzed in detail the effect of coverage, temperature, and surface reconstruction on the NO adsorption on Pt(110). The spectra resulting from the adsorption of NO on Pt(110) at 300 K (no (1 × 2) reconstruction) at high NO exposure (nearly saturation coverage corresponding to ca. 1 monolayer (ML)) present three characteristic features: a first strong broad band centered at 1796 cm-1; a second band centered at 1712 cm-1; and a third weak band at ca. 1613 cm-1.29 The first and third bands were attributed to N-O stretching vibration (νNO) in atop and bridge-bonded NO species, respectively, while the band around 1712 cm-1 is believed to be either due to NO adsorbed in the long bridge site (across the top layer rows) or due to NO adsorption on defect sites of the (1 × 1) structure. Comparing now the UHV data with our own in situ spectra (Figure 3), the band between ca. 1800 and 1680 cm-1 must be related to atop NO, while the band at ca. 1600 cm-1 is assigned to νNO in bridged NO. Furthermore, the close resemblance of the infrared spectra recorded under electrochemical and UHV conditions indicates (i) that the surface structure is similar (well ordered (1 × 1) domains with little (1 × 2) defects) and (ii) that the NO saturation coverage should be similar in both cases, that is close to 1. This last observation is in agreement with our voltammetric estimate of the NO saturation coverage under electrochemical conditions, ca. 1 for experiments in perchloric acid.15 The above vibrational assignments for NO adsorption on Pt(110) under electrochemical conditions are also in good agreement with results reported by the Alicante group. For (near) saturation coverage, Gomez et al.13 observed a broad bipolar band at around 1740 cm-1. At lower coverage they observed another weak band at around 1590 cm-1. On the basis of a comparison with data from early vibrational studies under UHV conditions,30 the authors suggested that the band located at around 1740 cm-1 is related to atop NO, while the band at around 1590 cm-1 is related to bridged NO. Brown et al.29 also showed that at low coverages NO occupies the bridge site, while with increasing coverage a site switching from bridge to atop position occurs. The site switching was explained by larger repulsion between bridged NO than between atop NO molecules, which is believed to destabilize the bridge site in favor of the atop site at high coverages. A closer examination of the spectra, acquired in the potential region in which adsorbed NO is reduced (Figure 3), points to an apparent “intensity transfer” from the broad band located between ca. 1800 and 1680 cm-1 (atop NO) to the band around 1600 cm-1 (bridged NO), thus suggesting that the site-switching scenario is valid under electrochemical conditions as well. (29) Brown, W. A.; Sharma, R. K.; King, D. A. J. Phys. Chem. 1998, 102, 5303. (30) Gorte, R. J.; Gland, J. L. Surf. Sci. 1981, 102, 34.
Langmuir, Vol. 21, No. 4, 2005 1451
Figure 4. The potential difference infrared spectra acquired at a Pt(110) electrode covered with NO adlayers at different coverage, as obtained by partial reductive stripping. All spectra were recorded at 0.60 V. Numbers in parentheses indicate the potential to which the potential was stepped prior to recording the spectrum at 0.60 V (see text for details). Experimental conditions: 0.1 M HClO4 (in D2O), reference potential 0.1 V, p-polarized light.
Accordingly, in the potential region between ca. 0.45 and 0.3 V the population of atop NO species decreases as a result of both reduction and site switching to bridge position. A more significant shift of the band between ca. 1800 and 1680 cm-1 toward lower frequencies is probably related to the combined effect of applied potential and decreasing NO coverage, since both factors are known to cause a decrease of the N-O (νNO) stretching frequency.18 At potentials around 0.3 V, atop NO is no longer detectable. A further decrease of the applied potential results in the disappearance of the band at ca. 1600 cm-1 and, accordingly, reduction of bridged NO. Similar trends with regard to the site occupation and site switching can be deduced by comparing the spectra acquired at different values of NO coverage at a potential where NO is stable. Figure 4 shows the potential difference spectra recorded at 0.6 V after a potential step to a given electrode potential, as indicated by the potential values in brackets. In this way, it was possible to separate the effect of the coverage from the effect of potential (at least for a given potential). A gradual decrease of NO coverage results in a gradual decrease in intensity of the band centered around 1760 cm-1 accompanied by the emerging (or more likely intensification) of the band centered around 1600 cm-1. The infrared data described here show that atop NO reacts in the potential region between ca. 0.45 and 0.3 V. Interestingly, a low-scan rate voltammetric profile presents the reduction peak E1110 in approximately the same potential window, irrespective of the supporting electrolyte (sulfate or perchlorate). Consequently, the voltammetric feature E1110 may be attributed to the reductive stripping of NO molecules bonded in atop position. An analogous
1452
Langmuir, Vol. 21, No. 4, 2005
Figure 5. Cyclic voltammograms for the reduction of NO adlayers at different initial coverage (as indicated by the numbers in the legend) on Pt(111) in 0.5 M H2SO4 at 2 mV s-1. The solid thick line is the voltammogram for reduction of a saturated NO adlayer, while the solid thin line is the Pt(111) blank voltammetry.
reasoning for the potential region between ca. 0.3 and 0.2 V implies that the voltammetric peak E2110 is related mainly to the reductive stripping of NO molecules adsorbed at bridge position. 3.2. Pt(111). The voltammetric profile for the reduction of a saturated NO adlayer on Pt(111) in 0.5 M sulfuric acid (Figure 5, thick line) shows three characteristic voltammetric features: E1111 at ca. 0.34 V, E2111 at ca. 0.28 V, and E3111 at ca. 0.15 V. The voltammetric profile is essentially the same in perchloric acid,15 proving the absence of any significant effect of the anion (co)adsorption. From Figure 5 it was deduced that the charge corresponding to the three features, but not their position, is a function of the initial coverage of NO. Interestingly, the voltammetric peak E3111 is “saturated” first with increasing coverage, thus questioning a scenario in which the three features would be due to different (consecutive) steps of NO reduction. Figure 6 shows that, quite similarly to reduction of NOads on Pt(110) (see Figure 2 and the accompanying explanations), the voltammetric features E1111, E2111, and E3111 can be reduced consecutively and virtually independently. The outcome is the same for experiments performed in perchloric acid. This result strongly suggests that the three voltammetric features are determined essentially by reduction of NO molecules at different adsorption sites and not by different processes. In contrast to NO adsorption on Pt(110), NO adsorption on Pt(111) under electrochemical conditions is quite well documented. The Alicante group has shown conclusively that stable NO adlayers can be generated on platinum, as well as other metal electrodes, either in situ (under potential control) or ex situ as a result of surface decomposition of nitrous acid.8,9,13 Their in situ infrared studies of NO adlayers on Pt(111), carried out mostly in the potential region where NO is chemically stable, revealed two spectroscopic features, ascribable to νNO in adsorbed NO: a bipolar (potential dependent) band at ca. 1430 cm-1, present at coverages as low as ca. 0.2 ML, and a second strong bipolar band centered at ca. 1680 cm-1, present at higher coverages.13 The higher frequency band was attributed to atop NO and the lower frequency band to bridged NO, based on early vibrational studies (see ref 13 and references therein). On the other hand, coverage
Rosca et al.
Figure 6. Cyclic voltammograms showing the independent stripping of the voltammetric features characteristic to NO reduction on Pt(111) (see text for details). Experimental conditions: 0.5 M H2SO4; 2 mV s-1. Numbers in the legend indicate the lower potential limit of the corresponding CVs. The solid thick line is the voltammogram for reduction of a saturated NO adlayer, while the solid thin line is the Pt(111) blank voltammetry.
dependent spectra, reported by Weaver et al.,18 show a pair of coverage dependent bands, centered at ca. 1440 and 1665 cm-1 at lower coverages of NO. With increasing coverage the latter band shifts to 1680 cm-1 and gains in intensity, while the former band disappears toward saturation. On the basis of the UHV data available at that time, particularly those questioning earlier vibrational assignments,31 Weaver et al.18 attributed the lower as well as the higher frequency band to NO located in the 3-fold-hollow site. However, more recent experimental and theoretical studies of NO adsorption on Pt(111), including scanning tunneling microscopy imaging,32 dynamic lowenergy electron diffraction analysis, and FTIRRAS UHV experiments,33 DFT calculations,34 and high-resolution X-ray photoelectron spectroscopy,35 have conclusively shown that at coverages up to ca. 0.25 ML the NO molecule occupies the face-centered cubic (fcc) 3-fold-hollow site, while at higher coverages (up to ca. 0.5 ML) the fcc 3-foldhollow NO species coexist with the atop NO species. Furthermore, it was suggested that at a coverage of 0.5 ML, that is the coverage close to the 0.4-0.45 ML range estimated for the NO saturation coverage under electrochemical conditions,9,15 adsorbed nitric oxide forms a phase consisting of one atop and one 3-fold-hollow-site NO per (2 × 2) Pt(111) unit cell. The recent vibrational assignments33,34 point to atop NO being related to the highfrequency band (ca. 1715 cm-1), while the low-frequency band (ca. 1490 cm-1) is attributed to NOads in the fcc 3-foldhollow site. The maximum coverage reported under UHV conditions (0.75 ML) is unlikely to be achieved under electrochemical conditions. Turning now to our own experimental infrared data, Figure 7 shows a set of the potential difference spectra for an (initially) saturated NO adlayer on Pt(111) electrode. (31) Brown, W. A.; King, D. A. J. Phys. Chem. B 2000, 104, 2578. (32) Matsumoto, M.; Fukutani, K.; Okano, T.; Miyake, K.; Shigekawa, H.; Kato, H.; Okuyama, H.; Kawai, M. Surf. Sci. 2000, 454-456, 101. (33) Matsumoto, M.; Tatsumi, T.; Fukutani, K.; Okano, T. Surf. Sci. 2002, 513, 485. (34) Aizawa, H.; Morikawa, Y.; Tsuneyuki, S.; Fukutani, K.; Ohno, T. Surf. Sci. 2002, 514, 394. (35) Zhu, J. F.; Kinne, M.; Fuhrmann, T.; Denecke, R.; Steinruck, H.-P. Surf. Sci. 2003, 529, 384.
Reduction of NO Adlayers on Pt
Figure 7. The potential difference infrared spectra acquired at a Pt(111) electrode covered with a preadsorbed NO adlayer. Numbers indicate the applied electrode potential on the RHE scale. Experimental conditions: 0.1 M HClO4 (in D2O), reference potential 0.1 V, p-polarized light.
In light of the above discussion, the potential-dependent band centered around 1680 cm-1 was attributed to νNO in the atop NO species. Under the conditions of the experiment shown in Figure 7, it is very difficult to detect and follow the evolution of the low frequency band corresponding to 3-fold-hollow NO, as this band is strongly masked by a broad feature between ca. 1400 and 1500 cm-1. This band is determined by a solution species, most probably a product of the reduction of adsorbed NO. Accordingly, this band is still present in spectra acquired using s-polarized light and it does not show any shift with applied potential. At the same time, this band gradually loses its intensity (corresponding to the accumulation of some species) in the potential region in which NOads is reduced.36 In the potential window in which NOads is stable, the peak frequency of the band located between 1700 and 1600 cm-1 (Figure 7) decreases linearly with decreasing the applied potential, with a “Stark-tuning” slope of ca. 75 cm-1 V-1. At lower potentials, a gradual decrease of the integrated intensity, broadening, and faster decrease of the band peak frequency with decreasing potential are observed, thus indicating consumption of atop NO species. At potentials around 0.3 V atop NO can no longer be detected. Nevertheless, the NO stripping voltammetry (36) Note that our in situ FTIR experiments were performed in D2O, although in the presence of a considerable concentration of protons (0.1 M HClO4 and H2SO4). The proton content was low enough to examine the 1500-1700 cm-1 frequency region, though apparently high enough to result in the formation of solution species, which may contribute in the frequency region between 1400 and 1500 cm-1. We identify these species as ammonium cations, containing H and D atoms in different ratios. Accordingly, theoretical calculations similar to those described elsewhere (see ref 40 for details) show that the species DNH3+, D2NH2+, and D3NH+ contribute around 1489 cm-1 (-NH3 (δs) mode), 1405 cm-1 (-NH2D (δs) mode), and 1457 cm-1 (-NHD scissoring mode), respectively.
Langmuir, Vol. 21, No. 4, 2005 1453
Figure 8. The potential difference infrared spectra acquired at a Pt(111) electrode covered with NO adlayers at different coverages, as obtained by partial reductive stripping. All spectra were recorded at 0.60 V. Numbers in parentheses indicate the potential to which the potential was stepped prior to recording the spectrum at 0.60 V. Experimental conditions: 0.1 M HClO4 (in D2O), reference potential 0.1 V, p-polarized light.
clearly shows that at this potential a significant amount of adsorbed NO (with coverage of ca. 0.2 ML) must be still present on the surface. Since no atop NO can be detected, the remaining NO molecules must be in the 3-fold-hollow sites, in accordance with the site occupation picture deduced for NO adsorption on Pt(111) under UHV and electrochemical conditions. Figure 8 shows results of an experiment similar to that described for Pt(110) (Figure 4), in which the potential was stepped back to the original potential of 0.6 V after each step to a lower potential. This experiment clearly indicates that in the potential window between ca. 0.45 and 0.3 V, atop NO species are stripped off the surface irreversibly, and do not exist in another (partially reduced) form. Comparing the voltammetric and spectroscopic data on NO adsorption and reduction on Pt(111), we conclude that atop NO is reduced in the potential window between ca. 0.45 and 0.3 V, which is the potential window in which the voltammetric peaks E1111 and E2111 are observed in the stripping profile. In other words, these two features are related to the reduction of NOads in atop position. The feature E3111 must then be caused by the reduction of NOads in the 3-fold-hollow site. Note that two voltammetric features (E1111 and E2110) correspond to the reduction of atop NO species. A possible explanation to this observation will be given in the Discussion. 3.3. The Product(s) of Reduction of Adsorbed NO. In discussing the products of NO reduction, it is important to make a clear distinction between continuous NO reduction and reduction of adsorbed NO. Continuous reduction of NO at polycrystalline platinum in acidic medium yields nitrous oxide (and molecular nitrogen) at
1454
Langmuir, Vol. 21, No. 4, 2005
Rosca et al.
moderately reductive potentials and hydroxylamine and ammonia at highly reductive potentials.14 Importantly, the presence of NO in solution is crucial for production of nitrous oxide (and molecular nitrogen).37 In absence of any NO in solution, formation of both ammonia and hydroxylamine is possible; however recent studies on the reduction of NO adlayers on polycrystalline and single-crystal platinum electrodes point to ammonia as the main product.9,13-15,38,39 A comparative analysis of NO adsorption on low-index single-crystal platinum indicates a remarkable concordance in the vibrational frequencies of saturated NO adlayers in electrochemical and UHV environments.18 Therefore, it would be reasonable to expect the adlayer structure and coverages to be similar on the corresponding surfaces in two cases, at least at room temperature (see also FTIR results on NO adsorption on Pt(110) in this study). Accordingly, an accurate analysis of the voltammetric charge gives coverages reasonably close to those under UHV conditions, provided that ammonia is assumed to be the (main) product of the reductive stripping.9,13,15,39 Finally, if any appreciable amount of hydroxylamine would be produced upon reduction of NO adlayers, unrealistically high saturation coverages would be implied (see ref 15 for a detailed discussion). Parts A and B of Figure 9 show potential difference spectra for (initially) saturated NO adlayers on Pt(111) and Pt(110) surfaces, respectively, acquired with spolarized light in water. The broad absorption band around 1460 cm-1 is determined by a species in solution and is clearly related to reduction of adsorbed NO. Thus, the intensity of this band is constant in the potential region in which NO adsorbate is stable and loses in intensity (formation of a product) in the potential region in which NO is reductively reduced on the corresponding surfaces. Both ammonia and hydroxylamine can contribute to this spectral region.40 However voltammetric results shown in Figures 1, 2, 5, and 6 clearly indicate almost complete stripping of NO adlayer and virtually complete recovery of the corresponding blank cyclic voltammograms (CVs), thus excluding formation of a significant amount of electrochemically active species, that is, hydroxylamine. Consequently, the absorption band centered around 1460 cm-1 was attributed to -(NH3) (δs) mode in NH4+.41 These results strongly suggest that ammonia is the main product of NO reduction on Pt(111) and Pt(110), presumably largely irrespective of the coverage. 4. Discussion The experimental data and considerations given in the previous section point to a prominent role of the NO adsorption mode in defining the voltammetric response for reductive stripping of nitric oxide adlayers. Accordingly, the voltammetric features observed for NOads reduction on Pt(111) and Pt(110) (and presumably those on stepped Pt[n(111) × (111)] surfaces as well) are determined essentially by reduction of species occupying different adsorption sites and not by different (consecutive) processes (reaction steps). Linearly bonded (atop) NO mol(37) de Vooys, A. C. A.; Koper, M. T. M.; van Santen, R. A.; van Veen, J. A. R. Electrochim. Acta 2001, 46, 923. (38) Gootzen, J. F. E.; van Hardeveld, R. M.; Visscher, W.; van Santen, R. A.; van Veen, J. A. R. Recl. Trav. Chim. Pays-Bas 1996, 115, 480. (39) Rodes, A.; Climent, V.; Orts, J. M.; Perez, J. M.; Aldaz, A. Electrochim. Acta 1998, 44, 1077. (40) Rosca, V.; Beltramo, G. L.; Koper, M. T. M. J. Phys. Chem. B 2004, 108, 8294. (41) Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. Spectrometric Identification of Organic Compounds; John Wiley & Sons: New York, 1991.
Figure 9. Potential difference infrared spectra acquired at a Pt(111) (A) and Pt(110) (B) electrode covered with a preadsorbed NO adlayer in the 1350-1900 wavenumber region. Numbers indicate the applied electrode potential on the RHE scale. Experimental conditions: 0.1 M HClO4 (in H2O), reference potential 0.1 V, s-polarized light.
ecules seem to be more reactive than multifold-coordinated NO molecules, whatever the surface orientation. The potential windows in which linearly-bonded NO, on one hand, and multifold-coordinated NO molecules, on the
Reduction of NO Adlayers on Pt
other hand, react on different surfaces are very much the same: compare the positions of features E1110, E1111, and E2111 (atop NO) and E2110 and E3111 (multifold-coordinated NO), Figures 1 and 5. It is important to emphasize that these observations alone do not allow deducing the actual reaction site. The actual reaction site may be an ensemble of sites, since the possibility of elementary steps, which may require additional adsorption site(s) (e.g., for the breaking of the N-O bond), cannot be ruled out. The voltammetric features E1111 and E2111 (Figure 5 and 6) were both ascribed to the reductive stripping of atop NO species, on the basis of the associated evolution of the FTIR spectra (section 3.2). Apparently, these two spectroscopically indiscernible (atop) species show some differences in their reactivity. This behavior may be tentatively ascribed to different local environmental of the atop NO species within well-ordered domains of NO adlayers (0.5 ML, 2 NO molecules per (2 × 2) unit cell)32-34 and atop NO species within somewhat disordered domains. Indeed, even under UHV conditions, the well-ordered phase corresponding to NO coverage of 0.5 ML requires special conditions (annealing of NO saturated surface at 225 K).33 Therefore, a highly ordered NO phase corresponding to 0.5 ML is unlikely to be achieved under electrochemical conditions, in which even more interfering factors may intervene. Interestingly, a comparison of NO reduction on Pt(111), Pt(110), and Pt[n(111) × (111)] stepped surfaces shows that NO adlayers are reduced in the same potential region between ca. 0.4 and 0.15 V, with little effect of coverage or anion adsorption. This observation, taken together with fact that NO reduction on Pt(100) takes place in the same potential region (reduction features between ca. 0.2 and 0.25 V),13 with little effect of coverage and anion coadsorption,42 suggests that differences in the surface structure of platinum electrocatalyst do not result in clear-cut differences in rate (overpotential) of NOads reduction. In other words, there is little structure sensitivity with regard to the (overall) kinetics of the process. The reduction of atop and bridged NO (features E1110, E2110, E1111, and E2111) proceed in the absence of any adsorbed hydrogen or in the presence of a very low coverage of coadsorbed hydrogen, thus confirming the possibility of the concerted electron-proton transfer steps (as expressed in eqs 1 and 2)). The experimental Tafel slopes for the voltammetric features E1110, E2110, E1111, and E2111 (Pt(110), Pt(111), and Pt[n(111) × (111)] surfaces) are all close to 40 mV decade-1, thus indicating the second electron transfer to be the rate-determining step (see the reaction schemes 1-3).15 On the other hand, the reduction of NO species adsorbed at the 3-fold-hollow site of (111) terraces (E3111) is characterized by the Tafel slope of 120 mV decade-1 and occurs in the presence of a considerable amount of coadsorbed H, thus raising the question of its role in the process. Indeed, it is not obvious that the reaction scheme (1-3) is valid in this case as well. It is noteworthy that the reductive stripping of NO adlayers on Pt(100) occurs in a potential region corresponding to high coverage of adsorbed H as well.13 Therefore, in the presence of adsorbed hydrogen, the reduction of adsorbed NO may, in principle, follow another mechanism that could involve catalytic (Langmuir-Hinshelwood-type) hydrogenation steps. Another factor contributing to the complexity of the issue of structure sensitivity of the electrochemical reduction of NOads is the mechanism of the N-O bond breaking. At the metal/vacuum interface, the (42) Rosca, V.; Koper, M. T. M. To be published.
Langmuir, Vol. 21, No. 4, 2005 1455
N-O bond breaking is a structure-sensitive reaction step requiring an ensemble of adsorption sites.43 This may be the case for NOads reduction at the metal/electrolyte interface as well. All together, at this stage it would be premature to conclude on the effect of the surface structure on the mechanism of NOads reduction on platinum. Despite the apparent structure insensitivity of the NOads reduction kinetics, the detailed voltammetric profile for NOads reduction is structure sensitive (Figures 1 and 5). Our results clearly indicate that NOads reactivity is differentiated by site occupation, which is influenced by the coverage of NO. However, at this stage it would be difficult to identify the molecular-level factors causing this differentiation in activity. Differences in adsorption strength would not provide a straightforward explanation (or may be even misleading) for differences in electroreduction activity of NO molecules occupying different adsorption sites. An important observation to be taken into consideration here is that the first two steps of the process are likely to be electrochemical steps (proton/ electron transfers),15 certainly for the reduction of atop NO species on the planes examined. This could mean that the adsorbate ability to take part in acid-base equilibrium (accompanying the electron transfer) and the stability of the resulting intermediates would determine the reactivity NOads species at different adsorption sites. The structure sensitivity of the NO reduction on platinum may be compared to the structure sensitivity of another extensively studied reaction, i.e., CO oxidation. Despite differences in the nature and mechanism of these processes, their structure sensitivity allows us to draw a number of interesting parallels. CO oxidation is a mechanistically structure-insensitive process. The LangmuirHinshelwood-type reaction between COads and OHads is commonly perceived as the rate-determining step, which is then followed by a decay of the resulting intermediate to CO2. In contrast to NO reduction, CO oxidation is a strongly structure-sensitive process with regard to the kinetics of the process. On Pt[n(111) × (111)] stepped surfaces, for instance, at moderately oxidative potentials, CO oxidation takes place preferentially at steps, which is related to the availability of OHads species, which are formed readily on steps.24-26,44 Due to its high mobility on (111) terraces, CO diffuses toward and reacts exclusively at steps (or defects in general). Under certain circumstances, the voltammetric profile of COads oxidative stripping may also exhibit several features. However, these features are not related to the oxidation of CO at specific adsorption sites.45 In contrast to CO oxidation, reduction of NOads proceeds at (or near) the site the molecule is initially confined to, this site being affected by the coverage; step sites and terrace sites are kinetically comparable. Furthermore, the site-differentiated activity in the case of NO reduction on platinum originates from the reactivity differences of linearly and multifold-coordinated NO species (and presumably the reactivity of the reduction intermediates as well), ultimately resulting in an adsorption-site-specific reaction and, consequently, an adsorption-site-specific voltammetric profile. 5. Conclusions The surface structure of the platinum electrocatalyst exerts a complex effect on the electrochemical reduction (43) Masel, R. I. Catal. Rev.-Sci. Eng. 1986, 28, 335. (44) Lebedeva, N. P.; Koper, M. T. M.; Feliu, J. M.; van Santen, R. A. J. Electroanal. Chem. 2002, 524-525, 242. (45) Markovic, N. M.; Grgur, B. N.; Lucas, C. A.; Ross, P. N. J. Phys. Chem. B 1999, 103, 487.
1456
Langmuir, Vol. 21, No. 4, 2005
of nitric oxide adlayers. The process is structure sensitive in terms of coverage-dependent site occupation. On Pt(111) and Pt(110) surfaces, linearly bonded NO species are more reactive than multifold-bonded NO species. On the basis of voltammetric and spectroscopic evidence, it was shown that different features observed in the voltammetric profile for the reductive stripping of NO adlayers on the two surfaces are related to the reduction of NO molecules at different adsorption sites and not to different (consecutive) processes. Furthermore, as deduced from spectroscopic and electrochemical data, ammonia is the main product of the reduction of adsorbed NO on Pt(111) and Pt(110). NO site occupation on Pt(110) under electrochemical conditions shows a remarkable accordance with the site occupation by NO under UHV conditions.
Rosca et al.
Accordingly, electrochemical reduction of the high-coverage NO adlayers (between ca. 0.5 and 1 ML) on Pt(110) is accompanied by a site switching from atop to bridge position, the bridged NO species being somewhat less reactive than atop NO species. For the electrochemical reduction of NO adlayers on Pt(111), atop NO species and the fcc 3-fold-hollow NO species coexist at high coverages (between ca. 0.25 and 0.5 ML) and may be reduced consecutively and independently. Acknowledgment. This work was supported by a grant from The Netherlands Organization for Scientific Research (NWO). LA0475831